The deep sea is the largest continuous ecosystem on earth, and plays a crucial role in global biochemical cycles, especially in the carbon cycle. At the same time, it remains one of the least explored environments. In the Arctic Ocean, changing environmental conditions at the ocean surface, e.g. decreasing sea ice coverage and changes in primary production, directly influence the deep sea through altered levels of carbon input. Considering that bacteria make up a major part of the deep-sea biomass and play an important role in carbon remineralization at the seafloor, it suggests itself that this taxonomic domain is one of the foci of deep-sea research. The present study was designed to get first insights into the response of bacterial deep-sea communities to different levels of carbon input in the form of chitin. This also included the exploration of the active community and related to this the development of an RNA extraction method. Therefore, we conducted a carbon enrichment experiment over 29 days using four different carbon (chitin) concentrations (0.001 – 1.0 mg carbon per ml sediment). Changes in community structure and activity were analyzed using cell counts, extracellular enzyme activity of chitin degrading enzymes, and T-RFLP community fingerprinting of the total and active community. The results showed a response of community structure and enzyme activity to carbon concentrations of 0.1 mg/ml sediment and higher. Community profiles differed between the total and active community structure, but patterns over time were similar for both fractions. Total community structure diverged with increasing carbon concentrations, whereas this trend was not apparent for the active community, which nonetheless showed clear differences between different time points. The community structure of the unfed controls showed unexpected shifts with time similar to the shifts observed for the lower carbon concentrated treatments, which might be due to the elevated experimental temperature. On average, 24% and 14% of the individual operational taxonomic units (OTU) showed linear correlations with time and extracellular enzymatic activity (EEA), respectively. Negative linear correlations were more common than positive ones, which might due to the broad phylogenetic distribution of chitin-degrading capabilities in the bacterial community, resulting in a large number of OTUs actively degrading chitin, but not correlating in significant positive linear relationships with time or EEA. In addition, relationships might be influenced by the fingerprinting technique, which yielded only relative abundances, so that changes in absolute numbers could not be assessed. Clear shifts in total and active community structure and activity could be evidenced over time for the different chitin concentrations, confirming certain aspects of other published and unpublished studies. Yet we lack comparative data regarding the response of the active bacterial community, since this was one of the first studies analyzing RNA from arctic deep-sea sediments. In the natural environment, carbon input occurs mainly with sinking particles. In order to get a first insight into pelagic-benthic links in bacterial diversity, we used environmental samples to compare community composition between the water column, sinking particles and the sediment surface at the HAUSGARTEN observatory. The comparison of the three oceanic fractions included the determination of bacterial abundances via cell counts, and an assessment of community structure using the fingerprinting technique automated ribosomal intergenic spacer analysis (ARISA), as well as 454 pyrosequencing. We were able to demonstrate the general dissimilarity of bacterial communities in the three fractions. In each environment several characteristic classes were identified, probably reflecting specific adaptations to the different environments. The deep-sea sediment was dominated by Gammaproteobacteria, Deltaproteobacteria, Actinobacteria and Alphaproteobacteria. The water column community was built up of Flavobacteria, Alphaproteobacteria, Gammaproteobacteria and Cyanobacteria, whereas the sinking particle-associated bacterial community of the sediment trap showed mostly Gammaproteobacteria, Flavobacteria and Alphaproteobacteria. Overall, this study provides further insights into the response of bacterial communities to changing levels of carbon (chitin) input, and contributes to a better understanding of community dynamics in response to environmental changes, which may in turn affect carbon cycling at the deep seafloor. In addition this work provides a starting point for future RNA-based studies in deep-sea surface sediments, e.g. metatranscriptomics.